🎉🎉🎉 Hey! If you’ve seen this post in the last couple of days, I have some updates for you at the end! 🎉🎉🎉
A couple of months ago, Geoff Boeing released a paper titled Urban Spatial Order: Street Network Orientation, Configuration, and Entropy. Basically, he looked at the directions that a city’s street network runs, and developed plots that display those orientations in a very clear and easy-to-understand format.
And because Dr. Boeing is awesome, he made the package (OSMnx) that makes all this happen open source, and better yet, put out a great set of examples, including the code that made the street network orientation graphs!
So of course I had to adapt it for some Metro Vancouver municipalities!
What you’re seeing here is basically counts of each street’s direction lumped into a plot. Take Port Coquitlam as an example. Most of its streets run north-south or east-west, but there are a few that run northwest-southeast.
There are a few obvious takeaways:
- Almost everybody is on a tight north-south/east-west grid.
- Belcarra is special because it’s small and its street network is super curvy.
- Welcome to New West, where everything is off by 45 degrees.
- A little bit of New West’s skew bleeds into Burnaby.
- Port Moody and West Vancouver both have sections that aren’t on a grid.
I didn’t get North Vancouver (either of them) in there because I couldn’t figure out how to get results for North Vancouver, and I’m not entirely sure if that’s actually Langley City or both Langleys lumped together. See below, I fixed all this (and it’s Langley City)!
The notebook is in my Azure Notebooks OSMnx Library (notebooks/17-street-network-orientations.ipynb), but at the time I wrote this blog post there was something hinky going on with Azure Notebooks so you might not be able to access it I was having issues with cookies that I’ve since resolved so you can probably get to them.
🎉🎉 Here’s the update! 🎉🎉
I figured out how to get the North Vancouvers and Langley Township into the analysis, and I also added Bowen Island and Lions Bay. Here are the updated plots, one for every municipality in Metro Vancouver:
I also got a few questions when I posted this page on Reddit:
Where did Kingsway go? The algorithm looks at every street in a city, and weights them equally. Kingsway is just one street, so it gets completely overwhelmed by every other N-S/E-W street in Vancouver or Burnaby.
Does this include lanes? I answered on Reddit that I didn’t know, but a comment from Jens von Bergman suggests that it does.
Do non-contiguous streets get counted twice? Yes, although not for the reason you might think. The algorithm looks at street segments, which are pieces of street between intersections and street ends, and not just at the total length of a street.
Does this factor in street length? Not this one, no! It counts each street segment the same, so if you had a street segment (remember, it’s the distance between intersections) that was a kilometer long, it would count the same as a street segment that’s only ten meters long. But the algorithm does have an option to weight street segments by length, so that kilometer-long segment would count 100 times as much as the ten meter long one. Here’s what the graphs look like for Burnaby, New Westminster, Surrey, and Vancouver for that scenario:
On Reddit I actually got the analysis of this wrong! This actually highlights the difference in block lengths between the different grid systems and has less to do with the physical dimensions of the city. In Surrey, the grid is very square. In Burnaby and Vancouver, blocks are longer along the E-W streets than they are on the N-S streets, and this is what shows up in the charts.
How come the Downtown core of Vancouver doesn’t contribute to a clockwise rotation? Because Vancouver’s Downtown (and West End) are a relatively small fraction of the total size of Vancouver! The tilted section of Vancouver on the downtown peninsula is roughly six square kilometers. The whole city of Vancouver is 115 square kilometers, so that downtown-ish area is about 5% of the whole city. If you zoom into the Vancouver graphs, the small tilted blip is about 5% the height of the major E-W/N-S spikes, so it all lines up as it should!